See the route at http://cycle.travel/map/journey/13541
In two places, the route description says "Continue on (unnamed route)" for a short section. Both cases correspond to a section of road that has an adjacent parking area, where the OSM data has the car-park boundary sharing nodes with the road. I presume C.T is routing through the car-park in preference to the road, and thus not finding a road name.
While I'm not sure the data is 100% correct (the car park doesn't go to the road centreline) it's probably a reasonable representation of the fact that the car-park is open to the road along a finite length. I think this representation is used fairly frequently in OSM, so probably C.T should probably try to handle this situation a bit better.
Comments
You’re absolutely right, and that’s pretty much the only thing I can hold against the wondrous OSRM, the piece of software which cycle.travel’s routing is based on. Sometimes if the same way connects two nodes, it chooses the wrong one: I’ve encountered exactly the same issue on the Pennine Cycleway (NCN 68) where it’ll suggest going through a car park for a short distance.
At the moment we’re running on an older version of OSRM (0.3.10) extensively patched for some of the things required for bike routing, notably route relations and elevation. Both of these are promised for OSRM trunk real soon now, and hopefully then we can get the shared nodes issue addressed.